Multiple Molecular Determinants for Allosteric Modulation of Alternatively Spliced AMPA Receptors

The Journal of Neuroscience, November 26, 2003 • 23(34):10953–10962 • 10953 Behavioral/Systems/Cognitive Multiple Molecular Determinants for Allosteric Modulation of Alternatively Spliced AMPA Receptors Jennifer C. Quirk and Eric S. Nisenbaum Neuroscience Division, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285 Positive allosteric regulation of glutamate AMPA receptors involves conformational changes that can attenuate receptor desensitization and enhance ion flux through the channel pore. Many allosteric modulators (e.g., cyclothiazide and aniracetam) preferentially affect the flip (i) or flop (o) alternatively spliced isoform of AMPA receptors, implicating residues in the flip–flop domain as critical determinants of splice variant sensitivity. Indeed, previous mutational analyses have demonstrated that the differential sensitivity to cyclothiazide and aniracetam depends on a single amino acid, Ser (flip) and Asn (flop), suggesting that this residue may be solely responsible for differences in modulation of AMPA receptor isoforms. The present studies tested this hypothesis by investigating the molecular determinants of modulation of AMPA receptor splice variants by a structurally distinct compound, LY404187, which displays strikingly different and opposing kinetics of allosteric regulation characterized by a time-dependent enhancement in potentiation of homomeric GluR1–GluR4i andatime-dependentreductioninpotentiationofGluR1–GluR4o.Site-directedmutagenesisofresiduesintheflip–flopdomainofGluR2 revealed that, although exchange of Asn775 for Ser in GluR2o was sufficient to confer the GluR2i phenotype of potentiation, the corre- sponding mutation, Ser775Asn, in GluR2i did not impart the GluR2o response. In fact, the GluR2o kinetics of modulation depended on a novel set of substitutions in GluR2i, including Thr765Asn, Pro766Ala, and Val779Leu in combination with Ser775Asn. Collectively, these results show that, unlike cyclothiazide and aniracetam, the residues that confer splice variant differences in modulation by LY404187 are not identical and indicate that allosteric regulation of AMPA receptors can arise from multiple molecular determinants. Key words: aniracetam; biarylpropylsulfonamide; cyclothiazide; flip; flop; alternative splicing Introduction tors can enhance performance of animals in models of cognitive The AMPA subtype of ionotropic glutamate receptors mediates function (Pontecorvo and Evans, 1985; Staubli et al., 1994b), as the majority of rapid excitatory synaptic transmission in the well as improve memory encoding in humans (Ingvar et al., brain. These receptors are tetrameric complexes assembled from 1997), suggesting that allosteric regulation AMPA receptors may subunits encoded by four different genes termed GluR1–GluR4 be a novel therapeutic approach for cognitive disorders (Yamada, (or GluRA–GluRD) in various stoichiometries, giving rise to 2000; Quirk and Nisenbaum, 2002). functional homomeric and heteromeric receptors (Hollmann et The preferential sensitivity of flip and flop receptors to allo- al., 1989; Keinanen et al., 1990). Additional diversity comes from steric regulation has focused efforts to elucidate the molecular alternative splicing of RNA for each subunit in a region that determinants of these differences on the flip–flop domain. In- encodes a 38 amino acid sequence in the extracellular domain deed, mutagenic studies have demonstrated that a single residue, between M3 and M4. These two splice variants named flip (i) and Ser750 in GluR1i and Asn750 in GluR1o, governs the greater flop (o) deviate in sequence by less than 10 amino acids and sensitivity of GluR1i to cyclothiazide and aniracetam, respec- display different rates of receptor desensitization that contribute tively (Partin et al., 1995, 1996). The critical nature of this residue to the heterogeneity of AMPA receptor synaptic responses in was further established in recent crystallography studies of cy- brain (Sommer et al., 1990; Dingledine et al., 1999). Flip and flop clothiazide in the ligand-bind core (LBC) of GluR2o containing isoforms also display preferential sensitivity to allosteric regula- this Asn to Ser mutation. Cyclothiazide was proposed to attenu- tion by distinct classes of compounds, including benzothiadia- ate desensitization by stabilizing the intradimer interface be- zines (e.g., cyclothiazide), biarylpropylsulfonamides (e.g., tween two adjacent ligand-binding cores, in part through binding LY404187), and pyrrolidinones (e.g., aniracetam). The func- to this Ser residue. In addition, when this GluR2o-Ser crystal was tional consequence of these positive modulators is to amplify superimposed with that of native GluR2o containing an Asn at excitatory synaptic transmission in part by reducing receptor de- this position, it was observed that tight binding of cyclothiazide sensitization (Dingledine et al., 1999). Significantly, an accumu- would be precluded (Sun et al., 2002). Collectively, these results lating body of evidence has demonstrated that positive modula- suggest that this single residue is responsible for conferring the sensitivity of flip and flop isoforms to allosteric regulation. Received July 30, 2003; revised Sept. 12, 2003; accepted Oct. 1, 2003. Recent studies with the biarylpropylsulfonamide compound Correspondence should be addressed to Dr. Eric S. Nisenbaum, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285. E-mail: [email protected]. LY404187 showed that human GluR2i were preferentially sensi- Copyright © 2003 Society for Neuroscience 0270-6474/03/2310953-10$15.00/0 tive to modulation compared with GluR2o (Ornstein et al., 2000; 10954 • J. Neurosci., November 26, 2003 • 23(34):10953–10962 Quirk and Nisenbaum • Allosteric Modulation of AMPA Receptor Isoforms Miu et al., 2001). The present studies have further demonstrated adjusted to 7.2 with KOH, and osmolarity was adjusted to 280 Ϯ 3 that the potency of LY404187 is greater for flip splice variants of mOsm/l with phosphocreatine). The extracellular solution contained (in all recombinant GluR1–GluR4, as well as revealed striking differ- mM): 120.0 Na isethionate, 5.0 BaCl2, 1.0 MgCl2, 20.0 CsCl, and 10.0 HEPES (pH was adjusted to 7.4 with 1.0 M NaOH, and osmolarity was ences in the kinetics of allosteric regulation of the two isoforms Ϯ for each subunit (see Fig. 1). To determine whether the Ser/Asn adjusted to 300 3 mOsm/l with glucose. Before recording, the transfected HEK cells were placed into a 50 mm residue is responsible for these splice variant differences, we used transparent plastic Petri dish that was mounted onto the stage of an site-directed mutagenesis to exchange residues in the flip–flop inverted microscope. After placing the recording electrode into the ex- domain of GluR2i/o. Our results show that, unlike cyclothiazide, tracellular solution bath, offset potentials were corrected, and electrode the molecular substrates for allosteric regulation by LY404187 of resistances ranged between 2 and 7 M⍀. Voltage-clamp recordings were AMPA receptor splice variants are not identical and do not de- made using an Axon Instruments (Union City, CA) 200B amplifier. The pend solely on the Ser/Asn residue. membrane potential of cells was held at Ϫ80 mV unless stated otherwise. Currents were digitized and monitored with pClamp software version 8.0 Materials and Methods (Axon Instruments) running on a Pentium personal computer. A small Molecular biology. Stable cell lines of HEK 293 cells transfected with amount of constant positive pressure (2–3cmH2O) was applied to the cDNA encoding human GluR1–GluR4 flip or GluR1, GluR2, and GluR4 electrodes as they were advanced through the bath. After achieving the flop subunits were kindly provided by Dr. Keith R. Jarvie (Allelix Biop- whole-cell configuration, series resistance was compensated (70–85%) harmaceuticals, Mississauga, Ontario, Canada). These cells lines were and monitored periodically. established as reported previously (Fletcher et al., 1995). HEK 293 cells Application of drugs was accomplished using one of two methods. For were transfected with the mammalian expression vectors pcDNA3.1 or concentration–response profiles, drugs were delivered using a 16-barrel ϳ ␮ pRc/CMV containing a GluR subunit gene. Transfected cells were se- pipette array made from small-diameter ( 600 m) glass capillary tub- lected on the basis of G418 resistance, and expression of genes was con- ing. Electronic valves controlled the gravity-induced flow of each solu- firmed by reverse transcription-PCR. The stable GluR3o cell line was tion from a 10 ml syringe to an individual barrel. The pipette array was ␮ established in a similar manner using AV12 cells. positioned 100–200 m from the cell before seal formation. The solu- ϳ For the mutagenesis and rapid perfusion studies, GluR2i and GluR2o tions from the drug array were changed ( 100 msec) by altering the were first subcloned into the mammalian expression vector pcDNA5/ array position with a DC actuator (Newport, Irvine, CA). For rapid per- FRT (catalog #V6010-20; Invitrogen, Carlsbad, CA). Native GluR2 sub- fusion experiments, a theta tube constructed from double-barrel boro- units are modified in the pore region (M2) so that a glutamine residue silicate glass tubing (catalog #TGC200-4; Warner Instruments, Hamden, (Q; CGA) is edited to an arginine (R; CGG). However, these edited CT) was used to deliver solutions to individual cells. The theta tubing was ϳ ␮ subunits have a significantly lower single channel conductance

Load more